Applied Energy 228 (2018) 426–436

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Economical flexibility options for integrating fluctuating wind energy in T

power systems: The case of China
⁎
Yi Dinga, , Changzheng Shaoa, Jinyue Yanb,c, Yonghua Songa,d, Chi Zhangb, Chuangxin Guoa
a
College of Electrical Engineering, Zhejiang University, Hangzhou 310058, China
b
Division of Energy Processes, KTH-Royal Institute of Technology, SE-100 44 Stockholm, Sweden
c
School of Business, Society and Engineering, Mälardalen University, SE-72123 Västerås, Sweden
d
Department of Electrical and Computer Engineering, University of Macau, Macau, China

H I GH L IG H T S

• The flexible resources for integrating wind power are analyzed from the cost perspective.
• The balancing cost when using coal-fired generation could reach $4/MWhw.
• The cost of DSM for integrating wind power can be very low.
• The EES will prevail coal-fired generation when its capital cost drops below 400$/kWh.

A R T I C LE I N FO A B S T R A C T

Keywords: The inherent stochastic nature of wind power requires additional flexibility during power system operation.
Wind power Traditionally, conventional generation is the only option to provide the required flexibility. However, the
Flexibility options provision of the flexibility from the conventional generation such as coal-fired generating units comes at the cost
Balancing cost of significantly additional fuel consumption and carbon emissions. Fortunately, with the development of the
Economical
technologies, energy storage and customer demand response would be able to compete with the conventional
generation in providing the flexibility. Give that power systems should deploy the most economic resources for
provision of the required operational flexibility, this paper presents a detailed analysis of the economic char-
acteristics of these key flexibility options. The concept of “balancing cost” is proposed to represent the cost of
utilizing the flexible resources to integrate the variable wind power. The key indicators are proposed respectively
for the different flexible resources to measure the balancing cost. Moreover, the optimization models are de-
veloped to evaluate the indicators to find out the balancing costs when utilizing different flexible resources. The
results illustrate that exploiting the potential of flexibility from demand side management is the preferred option
for integrating variable wind power when the penetration level is below 10%, preventing additional fuel con-
sumption and carbon emissions. However, it may require 8% of the customer demand to be flexible and
available. Moreover, although energy storage is currently relatively expensive, it is likely to prevail over con-
ventional generation by 2025 to 2030, when the capital cost of energy storage is projected to drop to ap-
proximately $400/kWh or lower.

1. Introduction anthropogenic carbon emissions, since electric power systems remain

The release of carbon dioxide (CO2) and other greenhouse gases
(GHGs) due to human activity results in a host of environmental issues
[1]. The enhanced public concern for adverse environmental impacts
associated with the use of conventional energy sources requires a
transition toward clean energy systems. Moreover, the de-carbonization
of electric power systems plays a significant role in reducing
the primary source of carbon emissions in the world. As a result, the
application of renewable energy in electric power systems generates
great interest. Among renewable energy sources, wind energy has ex-
perienced rapid development and has made significant inroads into
electrical power systems. Over the past decade, the global cumulative
installed capacity of wind energy has been growing at a rate of more
than 21% annually [2]. In 2015, global wind power capacity increased

⁎
Corresponding author.
E-mail address: yiding@zju.edu.cn (Y. Ding).

https://doi.org/10.1016/j.apenergy.2018.06.066
Received 3 January 2018; Received in revised form 9 June 2018; Accepted 11 June 2018
0306-2619/ © 2018 Elsevier Ltd. All rights reserved.
Y. Ding et al. Applied Energy 228 (2018) 426–436

Nomenclature Ii Status of the unit i (0 or 1)

ciu, cid Start-up/turn-off cost
Abbreviations Ru Ramp rate of the generating unit
Su Start-up ramp rate of the generating unit
CO2 Carbon dioxide Pigen Power out of the generating unit i
GHG Greenhouse gase Pldem Power demand of the electric load l
ESS Energy storage system Pwgen Power out of the wind plant w
DSM Demand side management Pwrate Rated capacity of the wind plant w
UC Unit commitment Pwpre
,t Day-ahead wind power forecast
ED Economic dispatch Pwavi, t Real-time wind power potential
SCCR System coal consumption rate rw Capacity credit of the wind power plant
CER carbon emission rate Vci/Vco Cut-in/cut-out wind speed
PCR ESS’s power capacity requirement Vr Rated wind speed
ECR ESS’s energy capacity requirement Ptdisc Discharging power of the ESS
LCR Load power capacity requirement Plinter Interrupted power of electric load l
avi
LCE Expected load-curtailment energy PESS PCR of the ESS (MW/MWw)
avi
MWhw Per MWh of wind energy production EESS ECR of the ESS (MWh/MWhw)
avi
MWw Per MW of installed wind power capacity PDSM LCR of the ESS (%)
avi
EDSM LCE of the ESS (MWh/MWhw)
Symbols ηc , ηd Charging, discharging efficiency
Δt Interval between two time periods
i, j Index of conventional generating units (subscript) pE Unit price of the ESS ($/MWh)
w Index of wind power plants (subscript) pDSM Unit price of the interrupted load ($/MWh)
day
l Index of electric load (subscript) ECGU Total conventional generation during a day
day
t Index of time periods (subscript) EWIND Total wind power generation during a day
k Index of the state ai , bi , ci Fuel consumption/carbon emission coefficients
NC Number of conventional generating units μ, σ Capacity credit of the wind power plant
NW Number of wind power plants σtotal Expected value and standard deviation of wind speed
NL Number electric loads σwind, σload Standard deviation of the wind/load forecast error
NT Number electric loads

by 17.1%, from 369,705 MW to 432,883 MW. In China, wind power has
become the third largest power source, following thermal and hydro-
electric power, and generates 4% of the country’s electricity [3].
However, the power generated from wind is fluctuating and un-
certain [4], which presents significant challenges to the efficient utili-
zation of this energy source [5]. As electricity demand and supply must
be maintained in balance at all times, power systems need to absorb the
electricity fluctuation from wind power. An increasing capacity of
fluctuating wind power will increase the need for flexibility during
power system operation. Flexibility is the ability of the power system to
deploy its resources for re-balancing customer demand and generation
when fluctuations exist. For example, downward reserve is required to
ensure power system balance when the amount of injected wind power
is higher. Conversely, upward reserve is required when the amount of
wind power injection is lower. If there is not sufficient operational
flexibility, the efficient utilization of wind power cannot be achieved
[6]. The seriously wind power curtailment issue in China could well
prove that. The coal-dominated generation mix in China works against
the high level of wind penetration, since the flexibility of coal-fired
generating units is constrained by their ramp-up and ramp-down rates
as well as their minimum stable generation output. China’s inflexible
generation mix, which cannot respond well to changes in wind power
output, forces it to curtail a large amount of wind energy every year,
despite the country’s renewable energy ambitions. Wind energy cur-
tailment in China is becoming increasingly serious. The total energy
loss from wind curtailment from 2011 to 2015 was approximately 95.9
billion kWh [7], nearly equals to the gross electricity generated by wind
energy in Denmark in 2013 [8].
There is a general consensus that the intermittency and uncertainty
of wind power have been the major barriers for large scale wind power
integration. To deal with the uncertainty of wind power, many methods
have been developed to improve wind power forecasting accuracy
[9,10]. Moreover, many studies have been conducted on how to
improve the system flexibility so as to deal with the wind power in-
termittency, which usually combine the wind power with other flexible
resources [11–13]. There are three most important flexibility options
for integrating the fluctuating wind power, including using the oper-
ating reserves from the conventional generation, using the flexibility
provided by the energy storage or the demand-side management (DSM)
[14]. Traditionally, the flexibility from conventional thermal power
generation for providing operating reserves is the most important op-
tion for integrating fluctuating wind power. For example, the flexibility
for integrating the fluctuating wind power are typically provided by
conventional generating units in China, such as coal-fired power gen-
erating units [15]. However, utilizing the conventional generating units
to integrate the variable wind energy causes additional costs. Examples
include the short-term balancing services, provision of firm reserve
capacity, and more cycling and ramping of conventional plants for in-
tegrating the wind power [16]. Integrating the fluctuating wind power,
the conventional generating units should work at part-load and change
their output frequently to cope with the variability and uncertainties
associated with wind energy. Consequently, the operation of generating
units is varying and low load levels results in low energy efficiency,
higher fuel consumption and the additional cost. The additional cost
arising from the intermittency of wind and the subsequent causation of
‘balancing plants’ for system security is widely observed [17]. More-
over, the increased fuel consumptions come with the additional carbon
emissions. In other word, the effect of developing wind power in dec-
arbonizing the power systems is partially offset by the additional
carbon emissions due to providing required flexibility [18]. For-
tunately, with the development of smart grid technologies, energy
storage and DSM may be able to compete with the flexibility provided
by the conventional generation. Actually, energy storage and DSM
could be the preferred options since they avoid additional energy
consumption and emissions. Recent advances in electric energy storage
technologies provide an opportunity for using energy storage to address

427
Y. Ding et al.
the wind energy intermittency. There are already some investigations
on the use of energy storage applied to wind turbines for buffering the
variability of the output. It is convinced that the energy storage system
(ESS) is able to reduce the variability and uncertainty of short-term
wind power [19], and lift the capacity credit of wind power [20], and
increase the profit of the wind power [21]. Meanwhile, DSM can be
another source of this required flexibility. Utilization of smart grid
technologies in power systems creates opportunities to more efficiently
balance supply and demand [13,22]. It has been demonstrated that
traditionally passive loads may become a resource that can mitigate the
consequences of wind’s variability [23]. Other studies have found that
utilizing the DSM is able to compensate the wind power forecasting
uncertainty as well as reduce the total operational cost and air pollutant
emissions [24].
Providing the flexibility and balancing power to integrate the
variable wind power results in additional cost, no matter which kind of
flexible resources is used. There exists some studies analyzing and
evaluating that cost which is referred to as the “balancing cost” or
“integration cost” [16,25–28]. The balancing cost is the cost of the
flexible resources for integrating the wind power, expressed in dollars
per megawatt-hour of wind power generation ($/MWhw). Moreover,
most studies analyzed the balancing cost of wind power assuming that
the flexibility is provided by the existing generation resources. To the
best knowledge of the authors, few studies have been conducted on
evaluating the balancing cost when utilizing different flexible resources
so as to find the most cost-effective way to integrate the variable wind
power. Since the increasing penetration of wind power is redefining the
requirement for flexibility, it is necessary to provide enough operational
flexibility in the most economic manner. Therefore, this paper evaluates
the balancing cost when utilizing different flexibility options in order to
find which the best option is from a cost perspective. In this way,
guidance for investment in these flexible resources can be provided in
this paper for guaranteeing a pre-determined wind power development
plan.
The research idea and contributions of this paper can be summar-
ized as Fig. 1. Firstly, a multi-state wind power model is developed to
represent its variability and uncertainty. Then, the key indicators are
proposed respectively for the different flexible resources to measure the
Fig. 1. Research idea and contributions of the study.
428
Applied Energy 228 (2018) 426–436
balancing cost. Finally, the optimization models are developed to
evaluate the indicators to find out the balancing cost when utilizing
different flexible resources.
(1) A multi-objective stochastic economic dispatch (MSED) is proposed
to minimize the system operation cost and carbon emissions. Based
on the MSED, the impact of the variable wind power on the con-
ventional generation’s energy efficiency of the can be assessed by
comparing the with-wind scenario and without-wind scenario. Such
negative impact is quantified using the system coal consumption
rate (SCCR) index. Moreover, the SCCR can be converted to the
carbon emission rate (CER). Consequently, the balancing cost when
utilizing the conventional generating units can be measured by the
additional fuel consumption and carbon emissions, which provides
a benchmark and allows comparison with other resources.
Moreover, the balancing cost also embodies the negative impact of
the wind power’s intermittency on the carbon emission reduction,
which is not considered in other studies.
(2) An optimization model is developed to determine the power capa-
city requirement (PCR) and the energy capacity requirement (ECR)
for integrating wind power using the ESS. The objective is to
minimize the ESS’s cost, including the capital cost and wear-out
cost. Based on those parameters, the balancing cost of using ESS to
integrate the wind power is also evaluated and compared with the
benchmark cost.
(3) A reliability-constrained wind-DSM combined dispatch model is
developed. Based on the model, the load power capacity require-
ment (LCR) and expected load-curtailment energy (LCE) of DSM
programs for integrating wind power are also determined. Based on
those parameters, the balancing cost of using DSM to integrate the
wind power is evaluated and compared with the benchmark cost.
A case study is conducted to find the characteristics of using dif-
ferent flexible resources to integrate wind power. Conclusions are given
in the Case Study section, which could provide guidance on the in-
vestment and future studies. It should be noted that this paper takes
China as an example for economically evaluating the key flexibility
options for satisfying the wind power development target because wind
Y. Ding et al. Applied Energy 228 (2018) 426–436

curve of a wind turbine [32].

⎧ 0, 0 ⩽ Vw, t ⩽ Vci
⎪ Pwrate ( A+ B× Vw, t + C× Vw2 , t ), Vci ⩽ Vw, t ⩽ Vr
Pwavi, t =
⎨ Pwrate, Vr ⩽ Vw, t ⩽ Vco
⎪
⎩ 0, Vco ⩽ Vw, t (3)

As shown in Eq. (3), the wind turbine starts generating at the cut-in
speed Vci, and is shut down at the cut-out speed Vco. Rated power Pwrate
can be obtained when the wind speed is between the rated speed Vr and
the cut-out speed Vco. Moreover, there is a nonlinear relationship be-
tween the power output and the wind speed when the wind speed lies
within the cut-in speed Vci and the rated speed Vr as shown in Eq. (3),
where the constants A, B, and C are presented in [32].
Fig. 2. Six-state wind speed model. As discussed above, the power output of the wind turbine can vary
continuously and intermittently from zero to the rated value depending
on the wind speed at the wind farm site. Wind turbines are therefore
power integration is especially significant in China [29]. However, the
usually represented by multistate models in analytical methods [33]. In
methods and models can be applied worldwide.
the multi-state wind generation model, the wind speed is represented
by a large number of discrete speed states. The model can be simplified
2. Methods to calculate the balancing costs when utilizing
by reducing the number of states at the cost of accuracy. It has been
different flexible resources
proved that the 6-state common wind speed model can be used for
reliability studies of power systems with reasonable accuracy [34].
This section describes the mathematical models, based on which the
Therefore, a simplified multistate power generation model for a wind
balancing costs of utilizing different flexible resources can be de-
power plant can be determined by combining the 6-state common wind
termined.
speed model with the power curve as shown in Eq. (3). The 6-state wind
speed model is shown in Fig. 2, in which the six wind speed states and
2.1. Mathematic model for evaluating the balancing cost of utilizing coal- the corresponding probabilities are given.
fired generating units Based on the 6-state wind speed model, the wind speeds for of a
wind power plant can be obtained from this model using (4)
2.1.1. Impact of wind power on the operation of the coal-fired generating
units For k = 1, 2, ⋯6
Traditionally, the generation portfolio was designed to provide ⎧Vw, t (k ) = μ + (k−3) × (5σ /3)
sufficient flexibility to cope with the variability and the forecast error of ⎨
⎩ Pr(k ) = Pr(Vw, t = Vw, t (k )) (4)
electric demand. In this case, the power balance can be formulated as:
where μ and σ are the expected and variance of the wind speed dis-
NC NL
tribution. Pr(k ) for the six states can be found in [34].
∑ Pigen dem
, t − ∑ Pl, t = 0, ∀ t
i=1 l=1 (1)
2.1.3. Optimal dispatch of the coal-wind system
where Pigen
,tdenotes the power output of generating unit i at time t, and As discussed above, a MSED is developed with different wind power
Pldem
,t denotes the electric demand of load l at time t . penetration levels to quantify the incremental coal consumption of the
As the generation capacity from wind power increases, the system generating units providing operating flexibility. The first objective
also needs to be able to cope with the variability and uncertainties function of the MSED is the overall system cost expressed in Eq. (5),
associated with the wind power [30,31]. In China's coal-dominated including the start-up cost and turn-off cost of the generating units, the
power system, the flexibility requirements are usually met through fuel consumption cost and the loss of load cost.
operating reserves provided by coal-consumption generating units. The
NT Nk
power balance in this case can be formulated as: ⎛ u d ⎞
minF 1 = ∑ ⎜ci, t + ci, t + ∑ Pr(k )·(fkfuel + f kloss ) ⎟
NC NL NW t=1 ⎝ k=1 ⎠ (5)
⎛ ⎞
∑ Pigen dem
, t −⎜∑ Pl, t − ∑ Pwgen
, t ⎟ = 0, ∀ t
c=1 ⎝ l=1 w=1 ⎠ (2) where
NT NC
where Pwgen (t ) denotes the power output of wind power unit w at time t ,
N NW fkfuel = ∑ ⎛⎜∑ (aigen + bigen Pigen
, t (k ) + ci
gen
(Pigen 2
, t (k )) )
, t − ∑w = 1 Pw, t is referred to as the “net load”.
gen
and ∑l =L1 Pldem t=1 ⎝ i=1
As shown in (2), the conventional generating units should work at
part-load and change their output frequently with the variation in ⎞
+ c wcur (Pwavi, t (k )−Pwgen
, t (k )) ⎟
customer demand as well as with the variability and uncertainties as- ⎠ (6)
sociated with wind energy. Moreover, the operation of generating units
NT NL
at varying and low load levels results in increased fuel consumption and
f kloss = ∑∑ rlloss (Plserve dem
, t −Pl, t )
associated carbon emissions. t=1 l=1 (7)
In (6), aigen , bigen , cigen
are the fuel consumption coefficients.
2.1.2. Multi-state wind power generation model
The second objective function is the expected carbon emission,
This section develops a multi-state wind power generation model to
which is denoted by Eq. (8)
describe the variability and uncertainty of wind power.
The generated power of a wind turbine varies with the wind speed NT Nk

at the wind farm site. The power output of a wind turbine can be de- minF 2 = ∑∑ Pr(k )·f kemission
t=1 k=1 (8)
termined from its power curve, which is a plot of output power against
wind speed. Eq. (3) is the mathematical expression for a typical power where

429
Y. Ding et al. Applied Energy 228 (2018) 426–436

NT NC
f kemission = ∑∑ (aiemission + biemission Pigen
, t (k ) + ci
emission
(Pigen 2
, t (k )) )
t=1 i=1 (9)
In (9), aiemission , biemission , ciemission
are the carbon emission coefficients.
Moreover, by introducing the social cost of carbon (SCC) [35], the
second objective function can be also represented by a social cost.
Therefore, the proposed MSED can be converted to a single-objective
optimization model.
The power balancing constraints are expressed as:
NC NW NL
∑ Pigen
, t (k ) + ∑ Pwgen dem
, t (k )− ∑ Pl, t = 0, ∀ k , ∀ t
i=1 w=1 l=1 (10)
The ramp-rate limits of the generating units are expressed as:

Pigen gen
, t (kt )−Pi, t − 1 (kt − 1) ⩽ Riu Ii, t − 1 + Siu (Ii, t −Ii, t − 1) ∀ t;kt , kt − 1
∈ {1, 2, k , ⋯Nk }
Pcgen gen d d
, t − 1 (kt − 1)−Pc, t (kt ) ⩽ Ri Ic, t − 1 + Si (Ii, t − 1−Ii, t ) ∀ t ;kt , kt − 1
Fig. 3. The projected capital cost of different energy storage technologies.
∈ {1, 2, k , ⋯Nk } (11)
The generation output limits for the conventional generating units need for coal-fired reserve generation and avoiding an increase in coal
are expressed as: consumption. Moreover, certain energy storage technologies are al-
ready cost-competitive with certain conventional alternatives. Other
Ii, t Pimin ⩽ Pigen
, t (k ) ⩽ Ii, t Pi
max
, ∀ t, ∀ k (12)
energy storage technologies are also close to being cost-competitive in
The generation output limits for the wind power are expressed as: other applications, and the costs are expected to decline in the coming
years. The projected costs of different storage technologies from the
0 ⩽ Pwgen avi
, t (k ) ⩽ Pw, t (k ), ∀ k (13) U.S. Energy Information Administration (EIA), Bloomberg New Energy
where Pwavi, t (k ) is calculated based on the above multi-state wind power Finance (BNEF) [39], AOE and financial advisory firm Lazard [40] are
model: shown in Fig. 3.
The cost projections of different storage technologies used in dif-
Pwavi, t (k ) = {fV − P (μ + (k−3) × (5σ /3))}, ∀ k (14) ferent areas, including renewable energy generation integration (REN)
The impact of wind power on the energy efficiency of coal-fired and application in a transmission system (Trans), are derived from the
power generating units is quantified using the SCCR and CER indexes. financial advisory firm Lazard. The solid symbols and hollow symbols
The SCCR of with-wind and without-wind scenarios can be obtained are the lower and upper bounds of the cost projections, respectively.
based on the two-stage dispatch model, and then converted to the CER. The lines represent the cost projections come from the Energy
Moreover, the additional carbon emissions can also be represented by a Information Administration (EIA), Bloomberg New Energy Finance
social cost. In this way, the total balancing cost of utilizing coal-fired (BNEF) and Navigant.
generating units to integrate wind power can be expressed as: With the help of energy storage, the variation of wind power, can be
smoothed out and the mismatch between the available renewable
fuel0
bal fuel emission (ΔSCCR/ SCCR0)· CGCU + ΔCER·SCC day power and the load can be addressed [41]. Considering energy storage
CCGU = ΔCCGU + ΔCCGU = day
ECGU
EWIND systems, the power balance can be formulated as:
(15) NC NW NL

fuel0 ∑ Pigen
,t + ∑ Pwgen
, t + Pt
disc
− ∑ Pldem
,t = 0, ∀ t
where SCCR0and refer to the SCCR and fuel cost of the coal-fired
CGCU i=1 w=1 l=1 (17)
generating units without wind power integration. ΔCER and SCC de-
where Ptdisc
denotes the discharging power of the ESS.
note the increase in carbon emissions and the social cost of carbon,
In this paper, an optimization model is developed to determine PCR
respectively.
day
In (15), ECGU day
and EWIND are the total power generation of the coal- and ECR for using ESS to integrate wind power. The PCR is expressed in
fired generating units and the wind power plants during the day, re- megawatts per megawatt of installed wind power. The ECR is expressed
day
spectively. ECGU day
and EWIND are calculated as: in, megawatt-hours per megawatt-hour of electricity production from
avi avi
wind power. The model to evaluate PCR (PESS ) and ECR (EESS ) is for-
NT NC NT NW
mulated as below.
day
ECGU = ∑∑ Pigen day
, t , ECGU = ∑ ∑ Pwgen
,t
(16) The goal is to minimize the ESS’s cost while making the wind-ESS
t=1 i=1 t=1 w=1
combined output to meet an hour-ahead predicted power output [42].
NT
2.2. Optimization model for sizing the ESS and determining the balancing total
minCESS = ∑ (ctcap + ctwear )
cost t=1 (18)
where ctcap
is the capital cost divided into the period t, ctwear is the ad-
In addition to using operating reserves from conventional gen-
ditional cost due to the rapid battery wear resulting from the deeper
erating units, wind power variability can be operationally mitigated
discharge in the wind-integration application [43].
using energy storage. In fact, energy storage appears to be an obvious
ctcap and ctwear can be expressed as:
option to deal with the variability of renewable sources and the un-
avi
predictability of their output [36,37]. In multiple application areas pE EESS ·Δt st + 1

around the world, energy storage systems (ESSs) have been deployed to
ctcap = N
∑y =Y 1
365
;ctwear = ∫s
t
w (s )ds
(1 + r ) y (19)
aid the integration of renewable energies, especially wind power [38].
Large-scale energy storage at the output of a wind farm can be used to In (19), pE is the per-unit cost of the ESS, expressed in $/MWh. w(s)
mitigate the variability and uncertainty of wind power, reducing the denotes the wear-out density function, expressed as [44]:

430
Y. Ding et al. Applied Energy 228 (2018) 426–436

pE b × (1−s )b − 1 ①. The probability of not having any generator trip while having an
w (s ) = avi
×
2× EESS × μ2 a (20) un-forecasted wind and load variation greater than the system re-
serve level. This scenario corresponds to the first term in (24).
In (20), s is the state of charge (SOC) of the ESS, a and b are the ②. The probability of having only one full generator trip and an un-
specific coefficients, μ is the ESS’s efficiency [45].The power balance forecasted wind and load variation greater that the system reserve
constraint is expressed in (17). level. This scenario corresponds to the second term in (24), which
The charging and dis-charging characteristics of the ESS can be corresponds to the probability of having a wind and load variation
expressed as: avi
greater than PDSM minus the power not available after the full
disc disc outage of generator i, Pnafoi, t .
⎧ EESS, t = EESS, t − 1 + ηc Pt ·Δt Pt ⩽ 0,charge ③. The probability of having only one partial generator trip and an
⎨ EESS, t = EESS, t − 1 + ηd Ptdisc ·Δt Ptdisc > 0,discharge unforecasted wind and load variation greater than PDSM avi
. This sce-
⎩
disc avi nario is similar to that in ② and corresponds to the third term in
⎧ 0 ⩽ Pt ⩽ PESS ∀ t ,discharge (24).
avi
⎨−PESS disc
⎩ ⩽ Pt ⩽ 0 ∀ t ,charge
N N
⎛ ⎞⎛ ⎞ P avi
PLSNOt = ⎜∏ (1−FOPi, t ) ⎟ ⎜∏ (1−POPi, h) ⎟ × ⎛⎜1−Φ ⎛⎜ DSM ⎞⎟ ⎞⎟
avi
EESS, t ⩽ EESS ,∀t (21)
total ⎝ i=1 ⎠ ⎝ i=1 ⎠ ⎝ ⎝ σtotal, t ⎠ ⎠
By running the above optimization model, the total ESS cost CESS
can be determined. Then, the balancing cost of utilizing energy storage ⎛ ⎞⎛ ⎞
N
to integrate the wind power is the total cost of installing and operating ⎜ N ⎟⎜ N ⎟
+ ∑ FOPi, t × ⎜ ∏ (1−FOPj, t ) ⎟ ⎜ ∏ (1−POPj, t )⎟
an energy storage project divided by the wind power system over its j=1
bal
i=1
⎜⎜ j ≠ i ⎟⎟ ⎜⎜ j = 1 ⎟⎟
life. The expressing of CESS is given by
⎝ ⎠⎝ j ≠ i ⎠
NY day avi N
EWIND ·365 ⎛ ⎛ PDSM −Pnafoi, t ⎞ ⎞
bal
CESS total
= CESS ∑ × ⎜1−Φ ⎜ ⎟⎟ + ∑ POPi, t
(1 + r ) y (22) σtotal, t i=1
y=1 ⎝ ⎝ ⎠⎠
In (22), NY represents the expected lifetime of the ESS. ⎛ ⎞
N N
⎛ ⎞⎜ ⎟
× ∏ (1−FOPj, t ) ⎜ ∏ (1−POPj, t )⎟
2.3. Optimization model for determining the balancing cost when utilizing ⎜ ⎟
⎝ j=1 ⎠ ⎜⎜ j = 1 ⎟⎟
the DSM
⎝j≠i ⎠
avi
With the development of DSM-enabled technologies, demand side ⎛ ⎛ PESS−Pnafoi, t ⎞ ⎞
× ⎜1−Φ ⎜ ⎟⎟
resources hold an untapped potential for increasing system flexibility ⎝ ⎝
σtotal, t
⎠⎠ (24)
and aiding the integration of fluctuating wind power [46,47]. Various
types of DSM programs have been implemented in China; interruptible where Φ(x ) denotes the normalized Gaussian distribution function of
load is one example. Interruptible load represents a consumer load that, the system forecast error and σtotal, t denotes the standard deviation of
in accordance with contractual arrangements, can be interrupted at the the total system forecast error. The load forecast error in time t can be
time of annual peak load by the action of the consumer at the direct modeled as a Gaussian stochastic fluctuating with a mean of zero and a
request of the system operator. While generators offer operational standard deviation of σl, t . Since it is assumed that both the load and
flexibility by providing the ability to increase their energy output, the wind power forecast errors are uncorrelated Gaussian stochastic vari-
load facilities, in contrast, provide the ability to reduce their energy ables, the standard deviation of the total system forecast error σtotal, t can
consumption to offer a reserve. Wind power uncertainty can be man- be given by
aged at a lower cost through this type of DSM program to address wind σtotal, t = 2 2
σwind , t + σload, t (25)
forecast errors. For example, when the real-time wind power output is
lower than the forecasted level, reduced energy consumption by a load The objective is to minimize the cost associated with the LCE while
facility addresses the demand and supply imbalances in the system. In guaranteeing PLSNO within the acceptable ranges.
this way, DSM provides the flexibility historically provided by coal- avi
minCDSM = pDSM EDSM (26)
fired generating units. The power balance considering the DSM can be
expressed as: PLSNOh ⩽ ϖ (27)
NC NW NL NL
The LCE is expressed as:
∑ Pigen
,t + ∑ Pwgen
,t + ∑ Plinter dem
, t − ∑ Pl, t = 0, ∀ t
i=1 w=1 l=1 l=1 (23) NT NL
avi
EDSM = ∑∑ Plinter
,t
To utilize DSM as the reserve for integrating wind power, it is ne- t=1 l=1 (28)
cessary to determine how much standby demand capacity is required.
In addition, the expected load-curtailment energy must be evaluated to After obtaining CDSM based on the above model, the balancing cost
determine the energy capacity value of flexible demand, to provide of utilizing the DSM to integrate the wind power is expressed as:
guidance on setting the interruptible tariffs that compensate consumers bal
CDSM day
= CDSM / EWIND (29)
for voluntary demand reductions. A quantitative model is developed to
determine LCR and LCE of DSM programs for integrating wind power
avi 3. Simulation results and analysis
with different levels of penetration. The LCR (PDSM ) is expressed in
percentage of the required flexible load among the total electric de-
avi
mand. The LCE (EDSM ) is expressed in, megawatt-hours per megawatt- 3.1. Parameters
hour of electricity production from wind power.
In the model, the DSM is introduced to mitigate the probability of There is no doubt that the simulation results, including the balan-
shedding load (PLSNO) arising from the variability and uncertainty of cing costs, depend on the wind power patterns. The variations of wind
electric demand and wind power. PLSNO comprises three components, occur on different time scales from seconds to seasons. Hence, the wind
as shown in (24) [48]. power data covering the entire year is necessary for obtaining the

431
Y. Ding et al. Applied Energy 228 (2018) 426–436

convincing results. In this paper, the daily wind power output profiles be concluded that on average, a 1-MWh production of wind power will
are obtained from actual historic data from the wind farms located in result in an increase of coal consumption of 13.68 to 17.64 kg de-
North China. Moreover, to cover the necessary information that de- pending on the wind power penetration level. Moreover, based on the
scribes the variability of wind power, 337 sets of daily wind power simulation results, it can be concluded that the balancing cost ranges
profiles are selected and used in this study. from $3.27/MWhw to $4.21 (20% wind).
The standard electric demand profiles derived from the historical
data are shown in Fig. 4.
3.2.2. Balancing cost of utilizing ESS to integrate wind power
In Fig. 4, Case1 to Case8 cover four seasons: spring (Case1 and Case
Since the ESS capacity requirements depend on wind power pat-
2), summer (Case3 and Case 4), autumn (Case5 and Case 6) and winter
terns, different simulation results are obtained on a daily basis. The
(Case7 and Case 8). The weekday and weekend scenarios are also se-
analysis results are shown in Fig. 8, which emphasizes the energy sto-
parated.
rage capacity requirements in 6% wind scenario and 20% wind sce-
In this paper, the coal-fired generating units are classified into three
nario.
types according to their capacities: 300 MW units, 600 MW units and
The results including the PCR, ECR and the balancing cost of uti-
1000 units. Based on the composition of the China’s coal-fired gen-
lizing ESS for integrating wind power are shown in Fig. 8. In Fig. 8, the
erating units in 2015, the shares of the three types generating units are
simulation results from each scenario are shown in the Histo-
set as 44.8%, 46.1% and 10.1%, respectively [49]. The coal consump-
gram + Probabilities graph. The “Counts” refers to the number of cases
tion curves of the generating units with 300 MW, 600 MW and
in which the requirements are in the range. The “Cumulative Counts”
1000 MW capacities are shown in Fig. 5 [50].
refers to the number of cases in which the requirements are lower than
the range, and the results are represented as the probabilities. As shown
3.2. Simulation results
in Fig. 8, with a wind penetration level of 6%, the ESS, PCR and ECR are
in most cases less than 0.45 MW and 0.3 MWh, respectively, for a 1-MW
3.2.1. Balancing cost of utilizing coal-fired generating units to integrate
wind power integration. When the wind penetration level increases to
wind power
20%, the PCR and ECR are 0.45 MW and 0.3 MWh, respectively.
In this section, the optimal system dispatch model is simulated with
The PCR and ECR and the corresponding balancing cost of using ESS
different wind power penetration levels to quantify the incremental
to integrate the wind power are summarized in Fig. 9. As shown in
coal consumption of the generating units providing flexibility for in-
Fig. 9, the capacity requirements are almost independent of the wind
tegrating wind power. Wind energy already generates 4% of China’s
power penetration level, which is an advantage of using energy storage
electricity and will expand its share to 6% before 2020. Moreover, on
to integrate wind power.
30 June 2015, China submitted its “Intended Nationally Determined
Contribution” (INDC), including the target to peak carbon emissions by
2030 at the latest, and increase the share of non-fossil energy carriers of 3.2.3. Balancing cost of utilizing ESS to integrate wind power
the total primary energy supply to approximately 20% by that time In addition to the LCR and LCE, the costs of utilizing DSM for in-
[51]. Therefore, five scenarios are developed with different wind power tegrating wind power are also shown in Fig. 10. The LCR is expressed in
penetration levels: no wind (reference case), 4% wind, 6% wind (2020 the share of the standby DSM-demand in the total electric demand. The
target), 10% wind (2030 target) and 20% wind (high wind power pe- LCE refers to the expected load curtailment. The costs both in lower cost
netration case). Moreover, both the incremental fuel consumption and scenario and higher cost scenario are shown in the top panel of the
carbon emissions are transformed into costs, which are further defined figure. It is suggested that companies in the tertiary sector reduced
as the balancing cost of wind power integration. demand in exchange for compensation of 1500 euro/MWh [52]. In this
The impact of wind power integration on the energy efficiency of paper, it assumes a load curtailment cost of $0.714 (5 yuan)/kWh
coal-fired power generating units is quantified using the SCCR index. (lower cost scenario) or $1.428(10 yuan)/kWh (higher cost scenario).
The SCCRs for the ‘‘4% wind’’ scenario, the ‘‘6% wind’’ scenario, the Analysis results show that in the 4% wind scenario, 5.54% of the
‘‘10% wind’’ scenario and the “20% wind” scenario are compared with total electric demand should be flexible and responsive for integrating
the reference scenario where there is no integration of wind power. wind power. However, the expected load curtailment energy LER is
Moreover, the SCCR can be converted to the CER, which denotes the fairly small, approximately 2.69 × 10−5 MWh per MWh of wind power
ton CO2 per megawatt-hour of electricity generation. electricity injection. Moreover, the expected cost is far below the bal-
The CERs of the different scenarios are shown in Fig. 6. The in- ancing cost of utilizing coal-fired generating units to integrate wind
tegration of wind power increases the carbon emission rate sig- power. The cost advantage of DSM is obvious until the wind power
nificantly. In the 4% wind and 6% wind scenarios, the carbon emission penetration level exceeds 10%. However, it requires at least 5.54% of
rates increase from 0.856 tCO2/MWh to 0.858 tCO2/MWh and 0.859 customers to participate in the DSM programs, as shown in Fig. 10. For
tCO2/MWh, respectively. The effect is more obvious when the share of
wind power integration is expanded to 10% and 20%. In 10% wind
scenario, the SCCR increases to 0.861 tCO2/MWh, while the carbon
emission rate further increases to 0.869 tCO2/MWh for the 20% wind
scenario.
The increased coal consumption and its associated carbon emissions
should not be ignored due to the predictable growth of wind power
integration. In the 6% wind scenario, which can be achieved by 2020,
the additional carbon emissions are estimated to be 15.56 million tons
per year. When the wind power penetration level reaches 10%, which
can be expected by 2030, the additional carbon emissions will further
increase to 26.87 million tons per year.
The influences of integrating fluctuating wind power on the coal
consumption rate, carbon emissions and balancing cost are summarized
as shown in Fig. 7.
Obviously, SCCR, CER and the balancing cost are influenced by and
are positively correlated with the wind power penetration level. It can Fig. 4. Electric demand profiles.

432
Y. Ding et al. Applied Energy 228 (2018) 426–436

Fig. 5. Coal consumption curves of generating units with different capacities.
10% wind power penetration, it requires 7.92% of customers to parti-
cipate in the DSM programs. Advanced metering infrastructure and the
availability of dynamic pricing to customers are necessary for reaching
these DSM participation levels [53]. When the penetration level in-
creases to 10%, the balancing cost of DSM increases dramatically and
makes it not a cost-effective option in higher wind power scenario.
3.3. Comparison and conclusion
To facilitate comparisons, the characteristics of the primary flexible
resources in integrating wind power are summarized in Table 1,
Fig. 7. The influence of integrating wind power on the coal consumption rate,
carbon emissions and balancing cost.
including the balancing cost, sensitivity to the penetration level of wind
power, and so on.
It is founded that the balancing cost for using coal-fired generating
units ranges from $3.27/MWhw to $4.22/MWhw. Moreover, utilizing
the conventional generating units to integrate wind power comes at the
cost of additional carbon emissions. Take China for example, the in-
creased carbon emissions due to utilizing coal-fired generating units to
integrate the clean but fluctuating wind power can be 15.56 million
tons per year. In other word, the effect of developing wind power is
partially offset by the additional carbon emissions from the generating
units providing integrating services.
Fortunately, the development of energy storage technologies and
DSM offers new sources of flexibility. The increase in fuel consumption

0.900
0.895 no-wind no-wind
Carbon emission rate (tCO2/ MWh)

0.890 4% wind 6% wind

0.885
0.880
0.875
0.870
0.865
0.860
0.855
0.850
0.845
0.840
0 40 80 120 160 200 240 280 320 0 40 80 120 160 200 240 280 320
0.900
0.895 no-wind no-wind
20% wind
Carbon emission rate (tCO2/ MWh)

0.890
10% wind
0.885
0.880
0.875
0.870
0.865
0.860
0.855
0.850
0.845
0.840
0 40 80 120 160 200 240 280 320 0 40 80 120 160 200 240 280 320
Number of simulation days Number of simulation days
Fig. 6. CERs for different scenarios, where 4%, 6%, 10% and 20% are the penetration levels of wind power.

433
Y. Ding et al.
Fig. 8. PCR, ECR of the ESS for the wind power integration.
and carbon emissions can be averted by applying these energy storage
and demand response technologies. Exploiting the potential of flexible
customer demand is the preferred option for integrating fluctuating
wind power when the penetration level is below 10%, which requires
7.92% of the customer demand to be flexible and available. Developing
energy storage technologies and DSM-enabling technologies, including
advanced metering infrastructure and dynamic pricing, are necessary
for utilizing DSM for integrating wind power. Moreover, as introduced
above, the balancing cost of utilizing DSM for integrating one-unit of
wind power is very sensitive to the wind power penetration level.
434
Applied Energy 228 (2018) 426–436
Despite the current relatively high cost, the advantage of energy
storage technology in avoiding incremental fuel consumption and
emissions is significant. Moreover, the ESS is likely to prevail over coal-
fired generating units by 2025, when the capital cost of energy storage
is projected to drop to approximately $400/kWh. Furthermore, the
balancing cost of utilizing ESS for integrating one-unit of wind power is
not dependent on the wind power penetration level, which makes it the
best option for providing the flexibility in a power system with high
wind power penetration.
Y. Ding et al. Applied Energy 228 (2018) 426–436

1.0 PCR ECR 1.0

ECR(MWh/MWhw)
PCR(MW/MWw)
0.5

0.5
0.0

-0.5
0.0
0 2 4 6 8 10 12 14 16 18 20

8
Higher cost Lower cost
Balancing cost ($/MWhw)

0
0 5 10 15 20
Wind level (%)
Fig. 9. PCR, ECR and balancing cost of an ESS for the integration of wind power.

4. Conclusions Table 1
The characteristics of different flexible resources in integrating wind power.
The ever-increasing wind power production poses great difficulties Flexible Balancing cost Depending on Additional When will be
in operating power systems and increases the requirement of opera- resources the wind level carbon cost-effective
tional flexibility. In addition to the operating reserves provided by emission
conventional generating units, the additional flexibility requirements
CGU $3.27-$4.22 Kind of Yes —————
can be fulfilled by ESS and DSM. Considering that the deployment of ESS Depending on No No The capital cost
the flexible resources will influence the integration of wind power the cost of ESS decreases to
technically and economically, this paper provides a systematic eva- $400/kWh
luation of primary key flexible resources. The characteristics of the DMS $0.5–$41 Very sensitive No Wind level
below 10%
primary flexible resources in integrating wind power are founded and
summarized, including their balancing costs and sensitivity to the wind-
power level. The finding can provide guidance on the investment of

Fig. 10. The LCR, LCE, and balancing cost of DSM programs for wind power integration.

435
Y. Ding et al.
those flexible resources to assist the wind power integration. Moreover,
the methods and models are expected to serve as references for the
future research in this field.
Acknowledgements
The research is supported by the National Key Research and
Development Program of China under Grant 017YFB0903400, and
national natural science foundation of China (NSFC) under Grant
51577167 and 51537010.
References
[1] Obama B. The irreversible momentum of clean energy. Science 2017;355. aam6284.
[2] Global wind report, annual market update 2015 (Global Wind Energy Council,
2016); http://www.gwec.net/wp-content/uploads/vip/GWEC-Global-Wind-2015-
Report_April-2016_22_04.pdf.
[3] Feng Y, Lin H, Ho SL, Yan J, Dong J, Fang S, et al. Overview of wind power gen-
eration in China: Status and development. Renew Sustain Energy Rev
2015;50:847–58.
[4] Jin T, Tian Z. Uncertainty analysis for wind energy production with dynamic power
curves. In: IEEE international conference on probabilistic methods applied to power
systems; 2010. p. 745–50.
[5] Slootweg JG, Kling WL. The impact of large scale wind power generation on power
system oscillations. Electr Power Syst Res 2003;67:9–20.
[6] Dong C, Qi Y, Dong W, Lu X, Liu T, Qian S. Decomposing driving factors for wind
curtailment under economic new normal in China. Appl Energy 2018;217:178–88.
[7] Data on the wind energy curtailment during 2011–2015. < http://www.
fenglifadian.com/news/201604/20732.html > 2016.4.
[8] Wind Power, Denmark, and the Island of Denmark. < http://euanmearns.com/
wind-power-denmark-and-the-island-of-denmark/ > ; 2015, 2.
[9] Feng C, Cui M, Hodge BM, Zhang J. A data-driven multi-model methodology with
deep feature selection for short-term wind forecasting. Appl Energy
2017;190:1245–57.
[10] Wang HZ, Li GQ, Wang GB, Peng JC, Jiang H, Liu YT. Deep learning based ensemble
approach for probabilistic wind power forecasting. Appl Energy 2017;188:56–70.
[11] Ju L, Tan Z, Yuan J, et al. A bi-level stochastic scheduling optimization model for a
virtual power plant connected to a wind–photovoltaic–energy storage system con-
sidering the uncertainty and demand response. Appl Energy 2016;171:184–99.
[12] Amrollahi MH, Bathaee SMT. Techno-economic optimization of hybrid photo-
voltaic/wind generation together with energy storage system in a stand-alone
micro-grid subjected to demand response. Appl Energy 2017;202:66–77.
[13] Jiang Y, Xu J, Sun Y, Wei C, Wang J, Ke D, et al. Day-ahead stochastic economic
dispatch of wind integrated power system considering demand response of re-
sidential hybrid energy system. Appl Energy 2017;190:1126–37.
[14] Ren G, Liu J, Wan J, Guo Y, Yu D, Yan J. Overview of wind power intermittency:
Impacts, measurements, and mitigation solutions. Appl Energy 2017;204:47–65.
[15] Kubik M, Coker P, Hunt C. The role of conventional generation in managing
variability. Energy Policy 2012;50:253–61.
[16] Hirth L, Ueckerdt F, Edenhofer O. Integration costs revisited – An economic fra-
mework for wind and solar variability. Renew Energy 2015;74:925–39.
[17] Simshauser Paul. The hidden costs of wind generation in a thermal power system:
what cost? Australian Econ Rev 2011;44:269–92.
[18] Ren G, Liu J, Wan J, Guo Y, Yu D. Overview of wind power intermittency: Impacts,
measurements, and mitigation solutions. Appl Energy 2017;204:47–65.
[19] Bludszuweit H, Dominguez-Navarro JA. A probabilistic method for energy storage
sizing based on wind power forecast uncertainty. IEEE Trans Power Syst
2011;26:1651–8.
[20] Holttinen H, Meibom P, Orths A, et al. Impacts of large amounts of wind power on
design and operation of power systems, results of IEA collaboration. Wind Energy
2011;14(2):179–92.
[21] Khalid M, Aguilera RP, Savkin AV, Agelidis VG. On maximizing profit of wind-
battery supported power station based on wind power and energy price forecasting.
Appl Energy 2017;211:764–73.
[22] Jonghe CD, Hobbs BF, Belmans R. Optimal generation mix with short-term demand
response and wind penetration. IEEE Trans Power Syst 2012;27:830–9.
436
Applied Energy 228 (2018) 426–436
[23] Broeer T, Fuller J, Tuffner F, Chassin D, Djilali N. Modeling framework and vali-
dation of a smart grid and demand response system for wind power integration.
Appl Energy 2014;113:199–207.
[24] Falsafi H, Zakariazadeh A, Jadid S. The role of demand response in single and multi-
objective wind-thermal generation scheduling: A stochastic programming. Energy
2014;64:853–67.
[25] Swinand GP, Godel M. Estimating the impact of wind generation on balancing costs
in the GB electricity markets. European: Energy Market; 2012. p. 1–8.
[26] Yang M, Bewley R. Integration of variable generation, cost-causation, and in-
tegration costs. Electricity J 2011;24:51–63.
[27] Yan J, Li F, Liu Y, Gu C. Novel cost model for balancing wind power forecasting
uncertainty. IEEE Trans Energy Convers 2017;32:318–29.
[28] Joos M, Staffell I. Short-term integration costs of variable renewable energy: Wind
curtailment and balancing in Britain and Germany. Renew Sustain Energy Rev
2018;86:45–65.
[29] Mahmoudi N, Saha TK, Eghbal M. Demand response application by strategic wind
power producers. IEEE Trans Power Syst 2015;31:1227–37.
[30] Ma J, Silva V, Belhomme R, Kirschen DS. Evaluating and planning flexibility in
sustainable power systems. IEEE Trans Sustain Energy 2013;4:1–11.
[31] Lu MS, Chang CL, Lee WJ, Wang L. Combining the wind power generation system
with energy storage equipment. Ind Appl IEEE Trans 2009;45:2109–15.
[32] Giorsetto P, Utsurogi KF. Development of a new procedure for reliability modeling
of wind turbine generators. Power Apparatus Syst IEEE Trans 1983:134–43.
[33] Cheng L, Liu M, Sun Y, Ding Y. A multi-state model for wind farms considering
operational outage probability. J Mod Power Syst Clean Energy 2013;1:177–85.
[34] Karki R, Hu P, Billinton R. A simplified wind power generation model for reliability
evaluation. IEEE Trans Energy Convers 2006;21:533–40.
[35] Nordhaus W. Estimates of the social cost of carbon: concepts and results from the
DICE-2013R model and alternative approaches. J Assoc Environ Resour Economists
2014;1:273–312.
[36] Su H-I, El Gamal A. Modeling and analysis of the role of energy storage for re-
newable integration: Power balancing. IEEE Trans Power Syst 2013;28:4109–17.
[37] Zhang N, Kang C, Kirschen DS, Xia Q, Xi W, Huang J, et al. Planning pumped
storage capacity for wind power integration. IEEE Trans Sustain Energy
2013;4:393–401.
[38] Wang K, Jiang K, Chung B, Ouchi T, Burke PJ, Boysen DA, et al. Lithium-antimony-
lead liquid metal battery for grid-level energy storage. Nature 2014;514:348–50.
[39] Zhao C, Wang Q, Wang J, Guan Y. Expected value and chance constrained sto-
chastic unit commitment ensuring wind power utilization. IEEE Trans Power Syst
2014;29:2696–705.
[40] Levelized Cost of Energy Analysis 9.0. < https://www.lazard.com/perspective/
levelized-cost-of-energy-analysis-90/ > 2015, 11.
[41] Elliott D. Renewable energy and sustainable futures. Futures 2000;32:261–74.
[42] Brekken TKA, Yokochi A, Jouanne AV, Yen ZZ, Hapke HM, Halamay DA. Optimal
energy storage sizing and control for wind power applications. IEEE Trans Sustain
Energy 2011;2:69–77.
[43] Zhou C, Qian K, Allan M, Zhou W. Modeling of the cost of EV battery wear due to
V2G application in power systems. IEEE Trans Energy Convers 2011;26:1041–50.
[44] Han S, Aki H, Han S. A practical battery wear model for electric vehicle charging
applications. Power Energy Soc General Meet 2013:1100–8.
[45] Choi Y, Kim H, Sciubba E. Optimal scheduling of energy storage system for self-
sustainable base station operation considering battery wear-out cost. Energies
2016;9:462.
[46] Zhong H, Xie L, Xia Q. Coupon incentive-based demand response: theory and case
study. IEEE Trans Power Syst 2013;28:1266–76.
[47] Cecati C, Ciancetta F, Siano P. A multilevel inverter for photovoltaic systems with
fuzzy logic control. IEEE Trans Ind Electron 2010;57:4115–25.
[48] Doherty R, Malley MO. A new approach to quantify reserve demand in systems with
significant installed wind capacity. IEEE Trans Power Syst 2005;20:587–95.
[49] The report on China’s Electric Power Reliability Indices, 2015 (China Electricity
Council, 2016). http://news.bjx.com.cn/html/20160720/753239.shtml.
[50] Liu F, Jiang X, Li Z. Investigation on affects of generator load on coal consumption
rate in fossil power plant. Power Syst Eng 2008.
[51] Sheikhi A, Bahrami S, Ranjbar AM. An autonomous demand response program for
electricity and natural gas networks in smart energy hubs. Energy 2015;89:490–9.
[52] Klobasa M. Analysis of demand response and wind integration in Germany's elec-
tricity market. Iet Renew Power Generation 2010;4:55–63.
[53] Ding Y, Pineda S, Nyeng P, Østergaard J, Larsen EM, Wu Q. Real-time market
concept architecture for EcoGrid EU—a prototype for European smart grids. Smart
Grid IEEE Trans 2013;4:2006–16.